Safe or pseudo mobility?

[Anonymous]

I would like to know if you do safe mobility (ie you only count safe squares for pieces, or squares that are well defended and not attacked by a piece of lower value than the one for which you are calculating mobility) or pseudo one (ie you count every empty square)

[Uri Blass]

I do pseudo mobility and I read in the past that somebody found that safe mobility is worse.

It is also logical that controlling squares is positive even if the piece cannot move to a square because the opponent may need a piece to defend the relevant square.

Uri

[smcracraft]

I do pseudo because a) it is faster b) I already had the code and
it was quicker and c) I can't afford to do a lot of stuff in terminal
node evaluation that has an approximate faster near-equivalent.

Stuart

[Alessandro Scotti]

Hi Andrea,
I do pseudo-mobility but I plan to experiment also with safe mobility in some situations, for example when a piece has little pseudo-mobility, in knight-vs-bishop imbalances and so on.

[Ron Murawski]

I do pseudo mobility. Safe mobility is too expensive for me because I lack an attack board. In the future I was going to try safe mobility, but after viewing this thread and the poll I may not.

Ron

[Tord Romstad]

If I could have voted twice, I would have chosen both options. Gothmog, my old engine, used a combination of pseudo-legal and safe mobility. Glaurung and Scatha, my new engines, use pseudo-legal mobility.

I would probably still have used safe mobility if I knew a cheap way to compute it without attack tables.

Tord

[Volker Böhm]

Hi,

Spike uses "some" safe mobility, currently under test.

I think you have to handle pieces differently.

Knights: Do not count positions attacked by pawns.
Bishops, Rooks: Do not count positions blocked by pawn defended pawns.
Qeens: Take only safe fields, stop if you found about 5.

Greetings Volker

[Vadim Bykov]

For evaulation mobility I use pseudo mobility, but for evaluation bad pieces i think to use safe mobility.